Filename decoder
Paste a model’s filename below. This tool breaks it into simple pieces so you know exactly what you’re about to download.
| Part | What it means |
|---|
How to read the result
Family is the model’s actual name — like knowing you bought a “Honda Civic” before you check the trim level. Parameters is roughly how big the model’s “brain” is — more usually means smarter, but also bigger and slower. Tuning tells you if it’s a base model (good at continuing text on its own) or an instruction-tuned model (built to chat back and forth with you). Quantization is how compressed the file is — like compressing a photo: a smaller file, a small loss in quality, and it’s the main thing deciding whether the model fits on your computer.
This tool makes its best guess. GGUF filenames don’t follow one strict, official format — just a habit most people follow. So an unusual or renamed file might only decode partially. When that happens, trust the parts it did find.
Why this exists
Nobody actually teaches you how to read a filename like
Q4_K_M — you either already know it, or you’re
guessing. This tool, and the
guide below, exist so you don’t have to guess.
See how Varven’s model picker works · Picking a model that fits your machine.
Common questions
Why didn’t it recognize my filename?
It knows the common, well-known patterns people use. A renamed file, a very new type of compression, or an unusual format might not decode fully — the tool tells you which parts it did figure out.
Does this tell me if a model will fit on my computer?
Not directly — it tells you what the file is, not whether it fits. Picking a model that fits your machine covers matching size to your hardware.
Is anything I type here sent anywhere?
No. Everything happens right in your browser — nothing is sent anywhere.